Policy sensitivity analysis: simple versus complex fishery models

Traditional behaviour sensitivity analysis is used to judge the sensitivity of model behaviour to uncertain assumptions about model formulations and parameter values. It is argued that it may be just as important to test the sensitivity of policy recommendations to uncertain assumptions. However, in complex dynamic models, policy recommendations typically have elements of randomness and possibly subjectivity. This problem is avoided by using optimisation to identify unique policies for different model assumptions. The case is a cod fishery for which a quota policy combined with a simple capacity policy is investigated. It is found that the policies are not very sensitive to the choice of biological model (aggregated or age classes), while they are highly sensitive to uncertain assumptions about nonlinear economic relationships. Interestingly, these findings seem at odds with the priorities in much modelling activity, be it in the field of biology or economics. Copyright © 2005 John Wiley & Sons, Ltd.